main.py

import os

import preprocessor as trim
from constants import *


def create_folder_structure():
    """Cria a estrutura de pastas
    necessaria, caso ela nao exista.
    Em seguida, faz a segmentacao.

    As pastas VALIDATION_OUTPUT, TEST_OUTPUT
    e TRAINING_OUTPUT contem os captchas 
    segmentados.
    """
    if not os.path.exists(TRAINING_OUTPUT):
        os.mkdir(TRAINING_OUTPUT)
    if not os.path.exists(VALIDATION_OUTPUT):
        os.mkdir(VALIDATION_OUTPUT)
    if not os.path.exists(TEST_OUTPUT):
        os.mkdir(TEST_OUTPUT)

    for f in os.listdir(TRAINING_FOLDER):
        filename = os.path.join(TRAINING_FOLDER, f)
        audios = trim.trim(filename, TRAINING_OUTPUT)
    for f in os.listdir(VALIDATION_FOLDER):
        filename = os.path.join(VALIDATION_FOLDER, f)
        audios = trim.trim(filename, VALIDATION_OUTPUT)
    for f in os.listdir(TEST_FOLDER):
        filename = os.path.join(TEST_FOLDER, f)
        audios = trim.trim(filename, TEST_OUTPUT)


if __name__ == "__main__":
    create_folder_structure()
    from model import train, test, individual_classifiers_results, get_final_model, final_test
    X_train, y_train, std_scale = train()
    captchas_svm, caracteres_svm, captchas_1nn, caracteres_1nn, captchas_3nn, caracteres_3nn, captchas_lda, caracteres_lda, elementos, corretos_svm,corretos_1nn, errados_svm,errados_1nn, corretos_lda,corretos_3nn, errados_lda,errados_3nn= individual_classifiers_results(X_train, y_train, std_scale)
    accuracy_captcha, accuracy_character, wrong, correct, elements = test(X_train, y_train, std_scale)
    print("Accuracia captchas: {}".format(accuracy_captcha))
    print("Accuracia caracteres: {}".format(accuracy_character))
    final_classifier, std_scale = get_final_model()
    final_test(std_scale, final_classifier)